Letter Cite This: Org. Lett. XXXX, XXX, XXX-XXX
pubs.acs.org/OrgLett
Asymmetric Synthesis of Functionalized trans-Cyclopropoxy Building Block for Grazoprevir Feng Xu,* Yong-Li Zhong,* Hongming Li, Ji Qi, Richard Desmond, Zhiguo J. Song, Jeonghan Park, Tao Wang, Matthew Truppo, Guy R. Humphrey, and Rebecca T. Ruck Department of Process Research and Development, MRL, Merck & Co., Inc., Rahway, New Jersey 07065, United States S Supporting Information *
ABSTRACT: A practical and asymmetric synthesis of a functionalized trans-cyclopropoxy building block for the preparation of the HCV NS3/4a protease inhibitor grazoprevir is reported. Intramolecular SN2 displacement−ring closure, followed by a Baeyer−Villiger oxidation, yields the desired trans-cyclopropanol with full control of diastereoselectivity. A terminal alkyne is then effectively installed using LiNH(CH2)2NEt2. Starting from (S)-epichlorohydrin, the cyclopropoxy building block is prepared in 51% overall yield with >99.8% optical purity without isolation of any intermediates.
T
Results from each of these strategies were unsatisfactory with our substrates, in terms of diastereo- and/or enantioselectivity, yield, and practicality for large-scale production. Instead of focusing on direct cyclopropanation for the preparation of cyclopropanols, we envisioned that a practical and stereospecific preparation of trans-2-substituted cyclopropanols could be realized from readily accessible starting materials 4 and 5a−b, requiring the following two key transformations: (1) asymmetric cyclopropanation by treating chiral epoxides 5a−b with phosphonate 4 through a [1,3]-phosphorus-Brook rearrangement7,8 and (2) stereospecific Baeyer−Villiger oxidation of cyclopropyl ketone 3,9 as depicted in Scheme 1.
he prevalence of chronic hepatitis C virus (HCV) infections worldwide remains extremely high, to the tune of approximately 130−150 million people.1 Recently, direct-acting antivirals have emerged to cure chronic HCV infection with high probabilities of success. In 2016, grazoprevir (1),2 a potent NS3/ 4a protease inhibitor, in combination with the HCV NS5a inhibitor elbasvir, was approved as the novel therapeutic Zepatier for the treatment of HCV by the FDA and EMA. Grazoprevir (1) possesses a unique trans-cyclopropoxy moiety with a carbamate linkage incorporated in the macrocyclic ring. The discovery of the function of the cyclopropoxy moiety was realized through extensive studies based on an understanding at the level of molecular structure about the binding interaction of the HCV virus with drug candidates.2 As a result, the fusion of a cyclopropane ring significantly improved the potency and selectivity against the genotype 1−3 NS3/4a protease enzymes as well as clinically relevant mutant enzymes2 but also raised the inherent challenge of synthesizing grazoprevir. To support the development of an effective synthesis of grazoprevir suitable for large-scale commercial production, we sought to realize a practical synthesis of building block 2 containing this unique cyclopropoxy moiety. The first generation synthesis3 of 2 relied on enzymatic resolution of a racemic Simmons−Smith cyclopropanation product followed by installation of the terminal alkyne through a lithium acetylide coupling. While effective, this process was deemed unsuitable for long-term use due to the hazards, inefficiencies of the process, and a low overall yield.3b Various methodologies have been developed to prepare substituted/functionalized cyclopropanes.4,5 However, methods for the practical asymmetric synthesis of cyclopropanols, especially 2-substituted cyclopropanols, are still lacking.4,5 To tackle this problem, we explored numerous approaches for the direct preparation of cyclopropanols including asymmetric Simmons−Smith cyclopropanation of enolized substrates,5h Kulinkovich reaction,5j,k Walsh’s cyclopropanation via methylene di-iodozinc,5c and 3-exotet ring-closing cyclopropanation.6 © XXXX American Chemical Society
Scheme 1. Grazoprevir and Retrosynthetic Analysis
This synthetic strategy and the structure itself were complicated by the presence of the requisite acetylene in building block 2. Introducing the acetylene functional group in precursor 3 (R2 = HCC(CH2)2) is unattractive and costineffective, requiring a multiple step synthesis of alkyne epoxide 5b.10 Alternative strategies using stepwise installation of the Received: September 14, 2017
A
DOI: 10.1021/acs.orglett.7b02867 Org. Lett. XXXX, XXX, XXX−XXX
Letter
Organic Letters
Further studies on the cyclopropanation supported the reaction pathway depicted in Scheme 2. In the presence of NaOt-Bu, epoxide 5a was formed rapidly at ambient temperature without any significant byproducts. The cyclopropanation proceeded slowly at ambient temperature, affording only about 22% of product 10 along with 25% of intermediate 8 (R = CONMe2) after 7 days. We confirmed that no ee erosion occurred during SN2 epoxide opening and subsequent intramolecular SN2 displacement−ring closure.15 As shown in Table 1, the presence of the sterically bulky dimethyl amide group15 proved beneficial in the diastereoselective intramolecular ring closure of 9 (R = CONMe2), affording the desired cyclopropane 10 in >200:1 trans/cis ratio. The resultant crude cyclopropyl amide 10 was treated directly with 1.8 equiv16 of MeMgCl in MeTHF at 55 °C to afford the desired methyl ketone 11 in 95% assay yield without any epimerization (Scheme 2). The Grignard addition to amide 10 resulted in a corresponding stable hemiaminal intermediate, as evidenced by slow hydrolysis upon aqueous NH4Cl quench and explaining the exquisite selectivity forming desired ketone 11 without any over addition of MeMgCl.17 Treatment of 11 with Py·HBr3 in MeCN afforded the desired dibromo ketone 12 in 96% assay yield.18 α-Bromination of the methyl group adjacent to the carbonyl group was fully suppressed in the presence of 0.5 equiv of pyridine. Several oxidants were examined to affect the necessary Baeyer−Villiger oxidation, with the strong oxidant, CF3CO3H, providing optimal reactivity. After optimization, we found that the Baeyer−Villiger oxidation was best performed by generating CF3CO3H in situ upon treatment of (CF3CO2)2O with urea− H2O2 complex (UHP)19 and 12 in EtOAc in one-pot at 0 °C, resulting in a mixture of the alcohol and acylated products 13 in 91% assay yield. The ratio of the mixture of components of 13 did not impact the performance of the subsequent synthesis. With Baeyer−Villiger products 13 secured, we next investigated installation of the terminal alkyne group under zipper reaction conditions. Interestingly, although the alkyne zipper reaction was discovered in 197520 and has been widely applied to prepare terminal alkynes,12 the evolution of the zipper reaction to a more practical process has received comparatively little attention.12 1,3-Diaminopropanide has remained the most commonly used base throughout this time frame.12,21 Treatment of 13 with lithium 1,3-diaminopropanide (Li-DAP) yielded the desired alkyne 14 in ∼80% assay yield (Table 2). However, LiDAP prepared by treating diaminopropane (DAP) with alkyl lithium in hexanes resulted in a thick triphasic slurry, which we viewed as a robustness risk toward implementation for large-scale
terminal alkyne group through C−C bond coupling do not improve the synthetic efficiency11 and may require a hazardous acetylide coupling process3b unsuitable for large-scale preparation. After thorough consideration of the retrosynthetic strategy, readily accessible alkene epoxide 5a emerged as our precursor of choice. As such, we envisioned that transformation of the terminal alkene to the desired alkyne functionality could be realized practically through a bromination−elimination sequence,12 while the use of the corresponding bromine masked acetylene precursor 3 would eliminate the incompatible reactivity of the unsaturated C−C bond under Baeyer−Villiger oxidation conditions. Treatment of (S)-epichlorohydrin (6) with commercially available 3-butenyl magnesium bromide in the presence of 1 mol % CuI afforded chlorohydrin 7 in 94% assay yield with nearly 100% regioselectivity.13 The most streamlined approach to the cyclopropyl methyl ketone 11 would involve the use of methyl ketone phosphonate 4 (R = COMe, Scheme 2). Employment of Scheme 2. Preparation of Dibromide 13
this strategy resulted in the desired ketone 11 in low yield (Table 1, entry 1), due to the competitive side reactions of the Table 1. Selected Results of Cyclopropanation with Chlorohydrin 7 and Phosphonate 4 entry
phosphonate 4
yield (%)
trans/cis
1 2 3 4
R = MeCO R = CN R = CO2Et R = CONMe2
45a 28a 71a 87c
>97:3b 3:1a >97:3b >200:1c
Table 2. Selected Results of Target-Driven Base Screening for Elimination of Dibromide 13
a
Determined by GC analysis.15 bDetermined by 1H NMR analysis. c Determined by HPLC analysis.15
corresponding methyl ketone intermediates under the basic reaction conditions. Given this incompatibility, a practical preparation of methyl ketone 11 was developed by using amide phosphonate 414 (R = CONMe2) to suppress side reactions during cyclopropanation. After workup, the crude chlorohydrin intermediate 7 was treated directly with phosphonate 4 in the presence of NaOt-Bu in MeTHF at 75− 80 °C to affect the desired cyclopropanation, furnishing the desired cyclopropyl amide 10 in 87% assay yield.
entry
basea
reaction mixture
yield (%)b
1 2 3 4 5
LiNH(CH2)2NH2 LiNH(CH2)2NMe2 LiNH(CH2)3NMe2 LiNH(CH2)2NEt2 LiNH(CH2)3NEt2
thick triphasic slurry thick slurry slurry homogeneous homogeneous
80% NA >90 >90 >90
a
Prepared by treating the corresponding amine in THF or MeTHF with n-BuLi. bDetermined by GC analysis.15
B
DOI: 10.1021/acs.orglett.7b02867 Org. Lett. XXXX, XXX, XXX−XXX
Letter
Organic Letters
Table 3. Scope of Elimination Using LiNH(CH2)2NEt2a
production. Indeed, about 5−10% of the corresponding bromoalkene intermediates were not consumed under the heterogeneous Li-DAP conditions. The use of other bases such as t-butoxide, LiNH2, LDA, LiHMDS, and DBU failed to provide improved results. In addition, we found that the use of sodium or potassium bases was not desirable, as these conditions led to significant decomposition of cyclopropanol 14 via a rearrangement to the corresponding aldehyde 15. At this point, it became clear that we needed to develop an alternative lithium amide base to overcome the heterogeneity issue such that the elimination could proceed robustly and reproducibly. Toward this end, we were particularly interested in exploring the reactivity and physical properties of LiNH(CH2)nNR2. We postulated that modification of the N-alkyl groups of these LiDAP analogs22 could be beneficial, allowing the elimination to be carried out in a homogeneous solution.23 Indeed, as N,Ndimethyl was replaced by N,N-diethyl, a homogeneous solution of the lithium amide was obtained (Table 2). Unlike the use of LiDAP, the elimination of 13 with LiNH(CH2)2NEt2 or LiNH(CH2)3NEt224 was then carried out in a homogeneous solution in MeTHF or THF, offering robust control of the reaction performance and resulting in near complete consumption of the corresponding bromoalkene intermediates (
